NIC Coil Project Finished

Finished the Twin Solid State Tesla Coils just a few minutes ago. Periodically working on the coils for over 8 months now I finally spent a good few hours to get the primary coils done. Everything is wired up and I have tested both coils both separately as well as together to see how they handle feedback.

H-Bridge went through two hardware revisions mostly due to cooling and space issues. The first design had an entirely different bridge which was level to the floor of the case. It used a linear regulator off a toriodial step down transformer. The second and final design used a Switch Mode Power Supply providing 12.7V which was broken up into other voltages with linear regulators to keep noise low. The control knobs also changed sides to bring them in reach of the power switch for operator safety. The driver circuit went through two hardware revisions. The first being a Duemilanove at 16MHz mounted beside a driver board. The second and final one being two stacked boards with the ATMEGA 328P running on the same PCB as the VCO at 20MHz.

The software went through five revisions in total. Rev 4 used a pulse timer system which worked but proved unreliable for audio input at frequencies above 1KHz and had a number of harmonics which it could not respond to. The final revision Rev 5 used a hybrid Zero Crossing Detector and Pulse Timing system which allowed up to 2.5KHz audio modulation while remaining completely stable and with no harmonic problems. Rev 5 also included reduced power consumption and higher noise immunity. The control system draws 30mA idle (mostly LED power consumption) to 50mA when fully active in Audio Input Mode.

Now all that is left is some tuning to get longer sparks, some cosmetic clean-up perhaps a washup and paint. To get a bit bigger sparks I plan to raise the primary coil up which should be easy as it was originally designed to do this. I reduced the primary winding to 20T from 21T which did almost nothing. The current draw of these things is 0.3A to 3.4A on the interrupter mode and 0.3A to 6.8A on the audio mode (0-2.5KHz).

That is all for now, documentation, schematics, and operators manual are next.

NIC SSTC NOV Update

 The North Island College, Solid State Tesla Coils are almost Done! Some adjustments and cosmetic things remain before they are finished. There is currently one bug left to iron out involving once-again gate driver IC shutdown! Grr they shutoff at random but no damage is done, it just becomes annoying to reset the power switch each time they latch-up! Other than that both units have been tested and do work, with the internal interrupter mode and audio modulation mode both functioning. Programming of the controller was done with the Arduino IDE and GCC AVR compiler. The spark-length is a little shorter than expected due to primary turns and placement of the primary coil form height. This will be adjusted when the other cosmetics are finished. This design turned out to be rather robust and despite numerous stress testing not a single MOSFET has died yet! I did blow one fuse by accidentally putting in the wrong value. Compare this to the last version of this coil which so far has "consumed" over 20 MOSFETS and 8 Fuses.

The details made the difference in this design! The top of the metal enclosure was slit to break circulating currents induced in the metal by the primary. In old designs this was a source of huge losses and heating, essentially the enclosure became induction heated. The diameter of the primary was increased and a PVC pipe was used as an insulator. The air gap combined with the PVC meant less heating issues and no arc-over from secondary to primary. The primary was also raised up to couple more to the coil and less to the base this further reduced losses and improved coupling despite the increased diameter. The redesign fixed allot of space and cooling issues. The H-Bridge no-longer gets hot only warm thanks to the cooling fan included and this is only under CW mode. In BPS everything is cold to the touch. The design changes significantly reduced losses! The coils are much more efficient than previous versions, they use about half the power to produce the same spark-length and most notably no heating issues in secondary or bridge.

Here's a quick view of the anterior driving circuitry in its finished form. Top is the H-Bridge Consisting of x4 IRFP460 MOSFETS and additional Blocking Diodes, Recovery Diodes and capacitors for DC blocking and bus smoothing. The Blue-Yellow transformer is the Gate Drive Transformer. To the right of that is the main power fuse in an inline holder at 10A 120V mains. Below the fuse is 2 large 1000uF 400V capacitors used in the voltage doubler circuit which has its power rectifiers tucked under right lip of the enclosure and beside the mains connection and power switch. Also tucked under the lip of the enclosure below the large smoothing capacitors is a small 5A 12V switch mode power supply for the driver and controller electronics. Below the H-bridge we see first the driver board with the yellow capacitor and below that is the controller which is a programmable microprocessor connected to knobs on control panel below and front of the enclosure. The one thing I am most proud of is incorporating a reprogrammable chip in this design. It has been very useful for troubleshooting and making changes without any physical changes. With the chip I could see on my computer - supply voltages, faults that have occurred in the driver, line noise etc and allowed me to tune the coil or operate it remotely. It was also very handy in producing precision pulses to control the coils output and modulate the power to limit it for overloads during high duty cycle operation. Currently noise is tripping off my fault detection routine to easily so some software changes are still needed before I release.

Variac Modification

A while back I purchased a Variac (Auto-Transformer) on eBay; for a rather good low price at the time. I had planned to replace or at-least supliment the cheap inbuilt voltage meter with something more useful and "cool looking". So I decided today was a good opportunity to do just that! and mounted my homemade volt/amp panel meter box to the Variacs output. The project turned out to not be that hard but did take a few hours due to finicky wiring, drilling holes, and hard to get to bolts.

Inside the variac you can see the core/case ground wire (Yellow/Green), Neutral Wire (White), Hot (Brown) and Tap Wire (Red). In the typical variable auto-transformer configuration.



Inside the panel a 20A Breaker / Switch, Analog volt-meter and outlet plug. The 50A shunt was added to accommodate the Amp-meter. The old volt-meter was originally wired to the output of the Variac but is now rewired to the input so I could see the mains voltage which is useful for troubleshooting other issues.

Here you see the panel meter box. It contains the two digital meters on the front (not shown) and 2 isolated switch-mode power supply units in the back (visible). The volt-meter runs off 12VDC and the current-meter runs of 5VDC, both power supplies must be isolated to measure both voltage and current without explosions! I like switch-mode supplies because they can run as low as 50VAC and up to 250VAC so the meters will work under different input supply voltages over a wide range. This variac was originally a 240V input 0-280VAC output so this allows me to still operate it in both European countries and in North-America without any issues (other than input meter being pinned to max).

Here are the results! Nothing high power to test out the output yet but the current meter is rated for continuous 50A I have noticed it can measure several hundred Amps for a short period of time however ;) . The volt-meter is a 700VAC true RMS panel meter both are from Sure Electronics on eBay. (Running a ~23W Compact Fluorescent Light seen below)

NIC Coil Update

Due to lack of planning on layout I have encountered leakage inductance problems in both the bridge and gate driver section. There are also overheating problems. This is a small set-back to finishing the NIC coils for the beginning of the program this year.

I have rebuilt the bridge circuit from scratch, it is much lower inductance layout and now 2 mosfets share 2 heatsinks rather than all 4 on one heatsink and all the blocking and recovery diodes on another. This has reduced the heating problem. The new bridge circuit was tested and the resonant frequency of the coils appears to be 240kHz with the current small top-load toriod lawn-mower wheel things. I will also include more capacitors in the voltage doulber supply and more across the MOSFETs power bus, this will likely mean explosive MOSFETs but will reduce the rate at which they explode ;). So far with the new layout testing at low power the MOSFET's now seem "indestructible" to upset driving conditions below resonant frequency and no problems ever above, likely due to the barrier diodes in series and recovery diodes that bypass the MOSFET body diode. In old designs with audio modulation this was always needed as I pulled the driver below resonant frequency during the off-state for playing notes. This load was far to capacitive for the slow MOSFET body diodes and always killed them. I may change to tuning above resonant frequency for audio modulation but this could stress the gate IC's.

It was found that the GDT had too many turns as I expected a lower resonant frequency around 100 - 200kHz, it seems the lowest I go with the current setup (coupling of primary and secondary, topload and secondary wire length) is from 220kHz to 260kHz , with 240kHz being about the average when placed over its own enclosure left standing in the middle of the room.
I plan to mount the GDT directly onto a plastic spacer over the heat-sink to make servicing the parts easier. I have much better driver IC's and I'm getting pretty close to perfect gate waveforms under load. I have remade the GDT to have only 10 turns made of computer networking cable with the 4 twisted wire-pairs.

I will be making bigger and lighter weight top-loads to replace the heavy foil covered rubber and metal wheels. This is to reduce the resonant frequency a bit more to reduce switching losses in the bridge and more so decrease heating in the gate driver IC's. The new IC's have arrived to replace the latchup problem with the 3710's I also have ordered many types of schottky diodes to help remove some dissipation from freewheeling current inside the driver IC's.

The coils enclosures are being rebuilt for neater and lower inductance layout and cooling needs. This will also mean a repaint to cover up scratchs and a new user interface / control panel.
Sorry no pictures I dont have time, plus no one really reads this blog :P